On this page
ChatGPT statistics for 2026 show a tool used by more than 1 billion people a week (OpenAI), where health is "one of the most common ways people use ChatGPT today" (OpenAI) and where patients now research and compare providers. This page collects the ChatGPT figures healthcare marketers need, from users and prompts to health use, provider choice, referral traffic, citations and accuracy, each with its source and method. Curve Compliance helps healthcare brands see and improve how they show up in ChatGPT and other AI answers. AI Visibility is a feature in Curve Compliance, and Curve also offers it as a managed service. The managed service is Curve AEO, with a team that handles the strategy and the work.
Want to know whether ChatGPT names your brand when patients ask about care like yours? Book a call to see AI Visibility on your own questions, or to talk about the managed service.
For the wider picture across Google and other assistants, see AI search statistics for healthcare marketers. Curve AEO tracks the same thing these studies point to: whether ChatGPT names and cites your brand.
Key takeaways
- ChatGPT passed 1 billion weekly users. OpenAI reported more than 900 million in March 2026 (OpenAI) and more than 1 billion in August 2026 (OpenAI).
- Health is a core use. More than 300 million people a week ask ChatGPT health-related questions (OpenAI).
- Most health chats happen outside clinic hours. OpenAI says seven in 10 healthcare conversations in ChatGPT in the U.S. happen outside of normal clinic hours, which it defines as 8am to 5pm local time (OpenAI).
- Patients use it to choose care. 36% of patients who searched for a new doctor cited AI tools like ChatGPT and Claude as an influence (rater8).
- It sends real traffic. ChatGPT drove 83.8% of AI referral traffic to health care websites (Conductor).
- It searches the web on about a third of prompts. 34.5% of queries in Semrush's U.S. clickstream data as of February 2026 (Semrush).
- Its citations rarely match Google's top 10. Across ChatGPT, Gemini, Copilot and Perplexity, only 12% of cited links ranked in Google's top 10 for the same prompt, on average (Ahrefs).
How many people use ChatGPT
OpenAI's own counts have climbed through 2026. These are company statements, global and not audited, and OpenAI has used both "weekly active users" and "regular users".
| When | What OpenAI reported | Source |
|---|---|---|
| July 2025 | More than 700 million weekly users sending more than 2.5 billion messages per day, about 29,000 per second | OpenAI and David Deming, NBER paper |
| January 2026 | More than 800 million regular users | OpenAI, AI as a Healthcare Ally |
| February 2026 | More than 900 million weekly active users and more than 50 million consumer subscribers | OpenAI |
| March 2026 | More than 900 million weekly active users; search usage nearly tripled in a year | OpenAI |
| August 2026 | More than 1 billion weekly active users | OpenAI |
OpenAI also says ChatGPT has "6x the monthly web visits and mobile sessions than the next largest AI app" (OpenAI). Outside data shows growth slowing in the U.S.: Semrush found ChatGPT's traffic "plateaued around November 2025, hovering near 1 billion monthly visits through early 2026," while the average number of queries per session jumped 50% to 1.75 by February 2026 (Semrush).
What it means for healthcare marketers: ChatGPT is now a mass channel, and people are asking it more per visit. Many roundups still quote older user counts, so check the date on any ChatGPT figure before you use it.
Who uses ChatGPT in the U.S.
Pew Research Center surveyed 5,123 U.S. adults in early 2025. Use is highest among younger and more educated adults, and 66% of Americans had not used it, including 20% who had heard nothing about it (Pew Research Center).
| Group (U.S. adults) | Have ever used ChatGPT | Source |
|---|---|---|
| All adults | 34% | Pew |
| Ages 18 to 29 | 58% | Pew |
| Ages 30 to 49 | 41% | Pew |
| Ages 50 to 64 | 25% | Pew |
| Ages 65 and older | 10% | Pew |
| Bachelor's degree only / postgraduate degree | 51% / 52% | Pew |
| High school or less | 18% | Pew |
The same survey found 28% of employed adults use ChatGPT for work, 26% of all adults have used it to learn something new and 22% for entertainment (Pew). OpenAI's research found that among users whose names could be classified as masculine or feminine, the share with typically feminine names had risen to more than half (52%) by July 2025, a sign that adoption now resembles the general adult population (OpenAI).
What it means: ChatGPT use skews younger and more educated, so its weight depends on who your patients are. West Health-Gallup found 55% of recent AI health users used AI built into web search, such as Google AI summaries (West Health-Gallup), so plan for both.
What people use ChatGPT for
| What was measured | Figure | Source |
|---|---|---|
| Messages that are "Asking" / "Doing" / "Expressing" | 49% / 40% / 11% | OpenAI |
| Consumer usage that is work-related / non-work | About 30% / about 70% | OpenAI |
| Conversations about practical guidance, seeking information and writing | Three-quarters | OpenAI |
| Messages classed as "Seeking Information" | Grew from 14% to 24% in the year to July 2025 | OpenAI, NBER paper |
| Prompts that matched no traditional search keyword | Between 65% and 85% for most of the study period | Semrush, U.S. clickstream |
| Prompts using traditional search language | Nearly doubled, from 18.9% to 34.9% (October 2025 to February 2026) | Semrush, U.S. clickstream |
| Queries where ChatGPT searched the web | 34.5% as of February 2026, down from 46% early in the study | Semrush, U.S. clickstream |
OpenAI's paper treats "Seeking Information" as a close substitute for web search, which is why its growth matters for anyone who depends on search traffic (OpenAI). Semrush notes that the share of search-enabled sessions has been volatile, from as low as 15% to as high as 66.3% (Semrush).
What it means: most prompts read like questions to an advisor, not keywords. Pages that answer a full patient question in plain language, with the details a patient would ask next, fit how ChatGPT is used.
ChatGPT and health
| What OpenAI reported | Figure | Source |
|---|---|---|
| Share of all ChatGPT messages about healthcare | More than 5%, globally | OpenAI, January 2026 |
| Regular users asking a healthcare question every week | One in four of more than 800 million | OpenAI, January 2026 |
| People bringing healthcare questions every day | More than 40 million | OpenAI, January 2026 |
| People asking health questions every week | Over 230 million in January 2026; more than 300 million in July 2026 | OpenAI; OpenAI |
| U.S. healthcare conversations outside normal clinic hours (8am to 5pm local time) | Seven in 10 | OpenAI, January 2026 |
| Messages per week about health insurance, U.S. | More than 1.5 million to nearly 2 million | OpenAI, January 2026 |
| Healthcare messages per week from underserved rural U.S. communities | Nearly 600,000 on average | OpenAI, January 2026 |
Health questions also come up inside ordinary chats. When OpenAI tested a dedicated health space, "more than 70% of health-related conversations took place outside the dedicated health experience" (OpenAI). U.S. surveys point the same way: 29% of adults use AI tools or chatbots like ChatGPT, Google Gemini or Claude for health information and advice monthly, up from 17% two years earlier (KFF), and among recent AI health users, 61% used general chatbots such as ChatGPT or Copilot (West Health-Gallup).
What it means: patients ask ChatGPT about symptoms, coverage and care outside clinic hours, when your front desk is closed. The answer they get then is your first impression.
ChatGPT and choosing a provider
In rater8's 2026 survey, 47% of patients had used an AI tool to research healthcare providers. Among patients who searched for a new doctor, AI tools like ChatGPT and Claude were cited by 36%, ahead of Google search results at 34% and doctor referrals at 32% (rater8).
Wolters Kluwer's 2026 survey of 254 U.S. patients points lower: only 30% would consider using AI to find a healthcare provider (Wolters Kluwer). The questions differ ("have used" versus "would consider"), rater8 sells reputation management software, and both are self-reported.
In rater8's 2025 report, patients who had used generative AI to research providers named ChatGPT most often as the most helpful or accurate tool (73%), followed by Google Gemini (56%), Microsoft Copilot (43%), Perplexity (30%) and Claude (17%); respondents could pick more than one (rater8 2025 report).
Errors are common and often unchecked: 66% of patients who used AI to research providers had encountered incorrect provider information from an AI tool, and 60% trusted the summary without verifying it (rater8). Across more than 350,000 locations of multi-location brands in all industries, SOCi found ChatGPT recommends just 1.2% of brand locations, compared with a 35.9% appearance rate in Google's local 3-Pack, "nearly 30 times more selective" (SOCi).
What it means: ChatGPT recommends only a small share of brand locations, so being one of them matters, and so does whether your details are right. See how to check whether ChatGPT names your practice and how patients use ChatGPT to find a doctor.
ChatGPT referral traffic
| What was measured | Figure | Scope | Source |
|---|---|---|---|
| Growth in ChatGPT outbound referral traffic | 206% in 2025 | U.S. clickstream, January 2025 vs January 2026 | Semrush |
| Share of ChatGPT referrals that go to Google | 21.6% | U.S., February 2026 | Semrush |
| Share going to the top 10 domains | Just over 30% | U.S., February 2026 | Semrush |
| ChatGPT's share of health care AI referral traffic | 83.8% (Perplexity 7.8%, Copilot 4.7%, Gemini 2.8%) | U.S., May to September 2025 | Conductor |
| AI referrals as a share of health care website traffic | 0.64% | Same dataset | Conductor |
| ChatGPT's share of all AI referral traffic | More than 80% | 81,947 sites, mid-2025 | Ahrefs |
| Google traffic vs ChatGPT, Perplexity and Copilot combined | 210 times more | Same dataset | Ahrefs |
Semrush says the Google share "could be that users go to Google to confirm their findings and navigate to brands they just discovered via ChatGPT" (Semrush). In other words, a ChatGPT mention can show up later as a branded Google search. See how to tie AI search traffic to bookings.
What ChatGPT cites
ChatGPT's sources often sit outside Google's first page. Ahrefs ran 15,000 prompts through ChatGPT, Gemini, Copilot and Perplexity: on average only 12% of cited links ranked in Google's top 10 for the same prompt, with Perplexity highest at 28.6% and the others around 8%, and 80% of citations did not rank anywhere in Google for the original query (Ahrefs).
Fresh pages get cited more. The average URL cited by AI assistants was 1,064 days old, against 1,432 days in Google's organic results, 25.7% fresher (Ahrefs). Citations also churn: tracking about 37,500 citations for 20 days, Ahrefs found 50% to 83% were gone the next day, though about 75% of those came back within 20 days (Ahrefs).
Brand signals off your site matter too. Across 75,000 brands, YouTube mentions showed the strongest correlation with AI visibility (about 0.737) in ChatGPT, AI Mode and AI Overviews, a correlation rather than proof of cause (Ahrefs).
What it means: judge ChatGPT visibility over several weeks, not from one check, and keep key pages current. See how to rank in ChatGPT for healthcare.
Do ChatGPT visitors convert?
Semrush found the average AI search visitor, "tracked to a non-Google search source like ChatGPT," is 4.4 times as valuable as a traditional organic visitor, based on conversion rate (Semrush). On Ahrefs' own site, AI search was 0.5% of traffic and 12.1% of signups, a 23 times higher conversion rate, mostly from ChatGPT (Ahrefs). Amsive's 54-site study is the counterweight: LLM referrals converted at 4.87% against 4.60% for organic search, a gap its paired test did not find significant (Amsive).
What it means: no study here measured patients. Follow ChatGPT visits to booked appointments on your own site before you count on any multiplier.
How accurate ChatGPT is on health questions
Research on ChatGPT models shows strong answers and real gaps. In a JAMA Internal Medicine study of 195 patient questions posted to a public forum, licensed health care professionals preferred the chatbot's answers to physicians' in 78.6% of 585 evaluations; the chatbot's answers averaged 211 words against 52 for physicians (Ayers et al.).
In JAMA Oncology, 34.3% of the cancer treatment outputs from an early version of ChatGPT (GPT-3.5) that made a recommendation also included at least one treatment not concordant with NCCN guidelines, and 12.5% of outputs included hallucinated treatments (Chen et al.). In Nature Communications, "even for GPT-4o with Web Search, approximately 30% of individual statements are unsupported" by the sources cited (Wu et al.). And in a Nature Medicine trial of 1,298 people, participants using GPT-4o and two other models identified the relevant condition in fewer than 34.5% of cases, no better than the control group (Bean et al.).
Model versions in these studies are older than today's. OpenAI itself says "ChatGPT can still make mistakes and does not replace the care and judgment of qualified medical professionals" (OpenAI).
What it means: clear, clinician-reviewed pages give ChatGPT accurate material to draw on, and give patients a reliable place to check what they were told.
How Curve Compliance helps
AI Visibility is a feature in Curve Compliance, and Curve also offers it as a managed service. In Curve, the feature is called AI Visibility (AEO), and the managed service is Curve AEO. Introducing Curve AEO: a service that helps you understand how your brand appears in AI search, with a team that handles the strategy and execution for you.
What You Get with Curve AEO:
- AI Visibility Tracking. See how often your brand appears in AI-generated answers.
- Optimize Your Site for AI. We make technical improvements that help AI assistants find, read, and understand your content.
- Competitive Insights. Understand who's getting recommended instead of you.
- Content Roadmap. Identify gaps and prioritize opportunities to improve performance.
With the managed service, Curve's team sets up your questions, competitors and topics, writes the pages on your roadmap, and makes the technical fixes that help AI assistants read your site. Your clinicians review and approve anything medical before it goes live. Each month, Curve's team walks you through your results on a call and sends a one-page report. Curve's team turns those insights into action: building your strategy, creating and optimizing content, and tracking progress.
It ties AI answers to the visits and bookings they bring you. It's built for healthcare: only the marketing question you track is sent to AI assistants, screened for PHI first, and never anything about a patient. Curve is SOC 2 Type II and signs a BAA on every plan.
In one AI Visibility program we've run, for a HIPAA-focused healthcare software brand, AI assistants named or cited the brand in more than half of their answers to its 40 core buyer questions about five weeks in. Across 184 buyer questions, its site was cited in more AI answers than any competitor's, more than 2.5 times as many as the nearest one.
Built for healthcare marketers, Curve connects the entire patient funnel in one platform: paid ads, web analytics, and now AI search. AI Visibility is available on request: book a call to get started.
Sources (checked November 6, 2026): OpenAI, expanding access to AI (August 31, 2026); OpenAI, next phase of AI (March 31, 2026); OpenAI, Scaling AI for everyone (February 2026); OpenAI and David Deming, How People Use ChatGPT, NBER (September 2025); OpenAI, How people are using ChatGPT; OpenAI, AI as a Healthcare Ally (January 2026); OpenAI, Introducing ChatGPT Health (January 2026); OpenAI, Launching Health in ChatGPT (July 2026); Pew Research Center, ChatGPT use (June 25, 2025); Semrush, ChatGPT traffic analysis (April 2026); KFF (June 17, 2026); West Health-Gallup (April 14, 2026); rater8, 2026 Patient Choice Report; rater8, 2025 report; Wolters Kluwer Health (2026); SOCi, 2026 Local Visibility Index; Conductor, Health Care 2026 AEO / GEO Benchmarks; Ahrefs, AI traffic growth (June 2025); Ahrefs, AI and Google overlap (August 2025); Ahrefs, fresher content (July 2025); Ahrefs, citation stability (October 29, 2026); Ahrefs, brand visibility factors (December 2025); Semrush, AI search traffic study (July 2025); Ahrefs, AI search conversions (June 2025); Amsive (September 2025); Ayers et al., JAMA Internal Medicine (June 2023); Chen et al., JAMA Oncology (August 2023); Wu et al., Nature Communications (April 2025); Bean et al., Nature Medicine (February 2026).
Frequently Asked Questions
How many people use ChatGPT?
OpenAI says ChatGPT serves more than 1 billion weekly active users, as of August 2026 (OpenAI).
How many people use ChatGPT for health questions?
OpenAI said in July 2026 that more than 300 million people a week turn to ChatGPT with health-related questions (OpenAI), and its January 2026 report says more than 5% of all ChatGPT messages globally are about healthcare (OpenAI).
How much traffic does ChatGPT send to websites?
In U.S. clickstream data, Semrush found ChatGPT's outbound referral traffic grew 206% in 2025 (Semrush). In health care, AI referrals were 0.64% of website traffic and ChatGPT drove 83.8% of them (Conductor).
Does ChatGPT search the web for every answer?
No. As of February 2026, ChatGPT enabled web search on 34.5% of queries, down from 46% early in Semrush's study of U.S. clickstream data (Semrush).
Do patients use ChatGPT to choose a doctor?
Many do: 36% of patients who searched for a new doctor said AI tools like ChatGPT and Claude influenced their choice (rater8). In rater8's 2025 report, 73% of patients who used AI for provider research named ChatGPT as most helpful or accurate (rater8).
How accurate is ChatGPT on health questions?
Results are mixed: evaluators preferred chatbot answers to physicians' in 78.6% of ratings in one study (Ayers et al.), while another found an early ChatGPT version mixed non-guideline cancer treatments into 34.3% of the outputs that made a recommendation (Chen et al.).
How do I know whether ChatGPT recommends my practice?
Ask ChatGPT the questions your patients ask, note which providers and sites it names, and check again over several weeks, since citations change from day to day (Ahrefs). Curve AEO helps you see how often your brand appears in AI-generated answers and understand who's getting recommended instead of you, with a team that handles the strategy and execution for you.
Talk to Curve Compliance
Want to see what ChatGPT tells patients about your brand? Book a call to see AI Visibility on your own questions, or to talk about the managed service.